A chatbot helps with some parts of studying and costs you on others, and here is how to tell which is which.
Nearly everybody I know uses a chatbot for school. Almost nobody talks about how. That silence is the actual problem.
There’s a version of this that makes you better at a subject. There’s another version that leaves you with a finished worksheet and an empty head. From the outside, those two evenings look the same.
So this isn’t a warning about technology. It’s about one way you get fooled. A clear explanation feels like understanding. Then the test asks you to produce the thing yourself, from a blank page.
Explaining a confusion in different words. This is the best thing it does. Your chemistry teacher explains limiting reagents one way. The textbook explains it the same way. You still don’t see it.
A chatbot will explain it as a sandwich shop that runs out of bread. Then as a table of moles. Then with numbers you pick. You can say “try again with a worked example” four times in a row. You won’t do that to a teacher with twenty-nine other students.
Making practice questions. This is the highest-value use, and almost nobody bothers with it. Paste in the topic list your teacher handed out. Ask for fifteen short-answer questions at the level of your course. Tell it to hold the answers until you ask. Writing questions is slow, which is why students re-read instead.
Checking your reasoning once you have some. Write the paragraph or work the problem yourself first. Then paste it in and ask where it would lose points. The order is the whole thing. Ask first and you’ve handed over the thinking. Ask second and you’ve got a second reader.
Turning a mess of notes into cards. Splitting three pages of notes into one-fact cards is mechanical work. Handing that over is fine. Just know what you give up. A lot of the value of writing flashcards happens while you write them. Rephrasing a definition is where you find out you didn’t follow it. Let it do the splitting, then reword every card yourself.
It doesn’t leave you stuck. It won’t let a sentence stay confusing for more than about eight seconds. Both of those are selling points. Both of them remove the exact thing that makes a memory stick.
This has been measured. In one study, some students re-read a passage and others tested themselves on it. The re-readers did better five minutes later. They did clearly worse two days later.
In another, spacing beat cramming for nine out of ten people. About seven in ten of them believed afterwards that the opposite had happened. You can’t feel this from the inside. The method that feels smoothest is usually doing the least.
Reading a good explanation is the smoothest thing in the world. You finish it nodding. Twenty minutes later you couldn’t reproduce it, and you have no idea. The feeling of having understood it is still sitting there, looking like knowledge. The same trap runs cramming.
Same tool, two completely different evenings. One sentence sits between them.
“Explain the causes of the Great Depression” gets you six tidy paragraphs. You’ll read them once and that’s it.
Now the other way to ask. “Ask me one question at a time about the causes of the Great Depression, wait for my answer before saying anything, then tell me what I left out.” That gets you a much worse forty minutes. It also gets you a grade you notice.
The test I use on myself is simple. Could I paste the reply straight into my notes without typing a sentence of my own? Then I didn’t study, I collected. Studying leaves a trail of your own wrong attempts behind it.
It has one tone, and that tone is certainty. A wrong answer arrives in the same calm voice as a right one. So confidence tells you nothing and you have to check. Every subject has a cheap check. Know yours.
In math, it will hand you an algebra step that looks clean and isn’t. Plug a number into the original expression and into the final one, and see if they match. In chemistry, it will balance an equation wrong and sound completely sure. Count the atoms on both sides.
In history, it invents quotes. It hands them to real people along with a page number that sounds specific. Check the name against the index of your own textbook. In Spanish, it writes sentences that are grammatical and that nobody says. You catch those against a text your teacher gave you.
There’s a second failure and it matters more. Push back and it folds. Type “are you sure?” after a correct answer and watch it apologise and hand you a wrong one. Agreement from a chatbot is not evidence in either direction.
Ask for a summary of a chapter you haven’t read and you get the vocabulary of that chapter. You get nothing underneath it. In class the next day you’ll recognise every term and feel fine. Then you can’t answer one question that uses them.
A summary compresses something you already have. It doesn’t stand in for having it. Reviews of study techniques rate summarising near the bottom of the list even when students write their own. One somebody else wrote sits further down still.
The version that works runs backwards from what everybody does. Read the chapter. Close it. Write four sentences on what it said. Then ask what you left out. Now it’s marking you instead of replacing you, and the four sentences you had to produce are what lasts.
Keep these somewhere you can paste them from. In every one of them, you’re the one producing sentences.
The fourth one is the most useful and the least comfortable. Writing out your own understanding is the moment you find out you don’t have one. That isn’t a reason to skip it. That’s the finding.
Plainly, with no speech attached. Using it to understand something is studying. Handing in text it produced as though you wrote it is cheating. Hard to detect doesn’t change what it is.
The middle is real and I won’t pretend otherwise. Getting it to fix your commas is one thing. So is getting it to tell you your third paragraph doesn’t follow from your second. That’s closer to what a writing centre does than to what a ghostwriter does. Where it lands depends on your teacher, and the rules differ by class, so ask instead of guessing.
The practical test is one question. Would you be fine with your teacher reading over your shoulder right then? You know the answer instantly, and it’s right more often than a policy document. The selfish version is simpler. You take the test alone. Anything you had it do for you in October is a thing you can’t do in May.
Use it to get unstuck. Use it to generate questions. Use it to check work you already did. Then turn whatever it helped you figure out into cards. Now test yourself on a schedule until you can produce it cold. More on that in how to write cards worth reviewing, and in how Machi.study decides when to show you one.